Digital Audio

[DTMark]

This is one for people into hi-fi.

That bit of me that's like the proverbial dog with the bone is manifesting again.

Digital Audio.

I've never been a fan of CD. It always sounds monstrously distorted at the peaks especially when there's a lot of high frequency information. This, to my ears, leads to a glassy, crunchy top end and the loss of essential timing information. It is why I have never been able to adopt CD. I can only listen for about 20 minutes and then my brain is so fatigued trying to correct information that it perceives is missing that I am too tired to carry on listening and have to turn it off.

The theory tells us that all we need is 16/44, in other words, CD.

Some so-called hi-fi purists stand by this theory and seem unwilling to entertain the idea that if you up the sampling and bit rate you will hear more information. Now clearly, 24 bit is higher fidelity than 16 bit. There's no debating that; mathematics dictates that this is so, and I don't think anyone is arguing against that. The only question is: can you hear the difference?

I'm told that people cannot reliably tell the difference between differing digital resolutions.

I can pick out 24/192 (hi-res) vs 16/44 (CD) every time with a 100% success rate be it on my computer, or through the main system with a modern DAC. It's dead easy. That horrible distortion at the top end and that awful choppy sense of truncation has gone. It's a total give-away.

And what continues to intrigue me is: why is it that I can do this, when many cannot? Why is it that most people can happily listen to 16/44 whereas I cannot tolerate it?

I have a couple of sample files. An original 24/192 mastered from an analog recording, and a downscaled 16/44 of the same. They are so obviously different that I am truly mystified as to why not everyone can hear this. I even had a friend into digital resample it various ways. I can always spot the 16/44 in a blind test within about five seconds.

I do have a number of theories as to why this is so and some are concerned with the way in which the brain processes ultrasonic information and the importance of it.

The sample files are copyrighted material that I don't want to link to here. It's just an album track from the early 1990s but the Moderators would be entirely right to take issue with me posting a link.

If you're intrigued: drop me a PM, I'll send you a link, and then tell me whether you can hear any difference and what it is. I've given some of it away but there's more.

I want to know why it is that I can hear something that others cannot.

[Frank Hovis]

You have a better "ear" by the um, sound, of it Mark.

Going back to the old portable radio cassette players I genuinely couldn't tell the difference between the mono and stereo versions of these so I realised early on that any money spent upon top end HiFi was money wasted.

I occasionally notice stereo when driving if it's fairly obvious that the drums seem to be coming from the passenger door but this seems more of a contrived trick rather than actually adding to the quality.

I notice my Roberts portable radio has a more realistic tone than my Pure one but that's about my limit.

[MrPin]

Well Mark I've replaced my 20 year old CD player with a modern one and even CDs sound good now. It does SACD too. Previously I had an ancient CD player which always sounded disappointing, and the "play everything" Blu Ray which would be a compromise with audio, although smashing with films.

[goldbug9999]

6 hours ago, DTMark said:

This is one for people into hi-fi.

That bit of me that's like the proverbial dog with the bone is manifesting again.

Digital Audio.

I've never been a fan of CD. It always sounds monstrously distorted at the peaks especially when there's a lot of high frequency information. This, to my ears, leads to a glassy, crunchy top end and the loss of essential timing information. It is why I have never been able to adopt CD. I can only listen for about 20 minutes and then my brain is so fatigued trying to correct information that it perceives is missing that I am too tired to carry on listening and have to turn it off.

The theory tells us that all we need is 16/44, in other words, CD.

Some so-called hi-fi purists stand by this theory and seem unwilling to entertain the idea that if you up the sampling and bit rate you will hear more information. Now clearly, 24 bit is higher fidelity than 16 bit. There's no debating that; mathematics dictates that this is so, and I don't think anyone is arguing against that. The only question is: can you hear the difference?

I'm told that people cannot reliably tell the difference between differing digital resolutions.

I can pick out 24/192 (hi-res) vs 16/44 (CD) every time with a 100% success rate be it on my computer, or through the main system with a modern DAC. It's dead easy. That horrible distortion at the top end and that awful choppy sense of truncation has gone. It's a total give-away.

And what continues to intrigue me is: why is it that I can do this, when many cannot? Why is it that most people can happily listen to 16/44 whereas I cannot tolerate it?

I have a couple of sample files. An original 24/192 mastered from an analog recording, and a downscaled 16/44 of the same. They are so obviously different that I am truly mystified as to why not everyone can hear this. I even had a friend into digital resample it various ways. I can always spot the 16/44 in a blind test within about five seconds.

 

Linear PCM encoding is a horridly inefficient format and 16 bits give way less dynamic range than the human ear, I'm not so convinced that higher sampling rates yield much benefit though. Have you tried down sampling to say 24/44 ?.

[Ash4781]

Back in the day when manufacturers made 14-16bit chips with resistor/ current source networks the digital filters that did all the oversampling weren't much cop. They had poor stop band attenuation for one so allowed noise through into the analogue stages. At the time it was to expensive to make the filters and the dac chips. From memory they all went down to massive oversampling and lower bits (1 bit oversampled 256X / delta sigma) being cheaper. 

info on DSP here http://www.dspguide.com/ch28/4.htm . Really do need some powerful hardware.

My understanding is the best stuff is all custom made where bods were writing their own filters, designing their own dacs,etc. Ed Meinter and Rob Watts are ones I read about. Problem is that stuff will be super expensive!

I tried non-oversampling dac but didn't like it!

[spyguy]

Once your older than 25 you might as well listen to music on a 30 year old Amstrad hifi.

Even now, my 10 yo goes  can you hear that high pitch whine. Err no.

[MrPin]

Mr Spy is as incimpthemsovle as always.

Some CDs are mixed really badly, and there's nothing that will improve that.

 

[dgul]

I think your problem is that you're listening to music.  Most people these days just have the music on and don't really care if it's 192 mp3 or something with more quality.    I can't stand hearing compression artefacts (particularly with piano -- dunno why), but most people don't seem to care.

FWIW, I do think that most digital audio is overly hacked about it, with too much (audio) compression to start with (wot Pin says).

Re hearing differences in the high quality formats -- This is possible.  I'd suggest that it would be more the bit depth over the sampling frequency.  But most people can't hear the difference in quality beyond 16/44 (so I guess the attitude would be something like 'we provide sufficiently high quality audio to please 99% of the population; there's no benefit for us in increasing costs/complexity to please that residual 1% and thus hit market share / technology adoption rates')

Part of the problem (for many people, not necessarily you) could be the actual sound of the hi-fi (as it were).  I gave up worrying about recorded sound quality when I started going to live performances (well, recitals, that sort of thing, not pop concerts) -- the actual sound of instruments playing is so far removed from that which you'd get when playing a recording.

[spyguy]

10 minutes ago, MrPin said:

Mr Spy is as incimpthemsovle as always.

Some CDs are mixed really badly, and there's nothing that will improve that.

 

Im FB messengering an ex at the mo. Crisis!

FB messenger messes with the text input!

[spyguy]

Youre are all precious flowers!

Turn up music till it sounds good!

Ditto for food - just put on more tomato ketchup til its edible!

[libspero]

Not sure... but then I can listen to MP3s provided they're more than about 160 joint stereo or 192 stereo.

Perhaps because that's practically all I've ever known.

I would have thought by the time you get to that quality it would be far more important how the original music had been recorded and what you were listening to it on.  

I probably couldn't tell the difference between 64 mono and perfect fidelity through a set of regular PC speakers 

[Craig_]

As a vinyl fiend I'd always been a bit sniffy about digital audio in general, but listening to Spotify on extreme quality (320kbps I think) through my B&W car stereo is probably the finest listening experience I've ever had. What does bug me with a lot of modern music is the absurd compression which is applied, makes everything sound like it was recorded in a toilet.

I realise this doesn't add much to the argument, my suspicion is that the OP has particularly highly-functioning ears, possibly honed by the ingestion of Class A's...

[stormymonday_2011]

12 hours ago, dgul said:

I think your problem is that you're listening to music.  Most people these days just have the music on and don't really care if it's 192 mp3 or something with more quality.    I can't stand hearing compression artefacts (particularly with piano -- dunno why), but most people don't seem to care.

FWIW, I do think that most digital audio is overly hacked about it, with too much (audio) compression to start with (wot Pin says).

Re hearing differences in the high quality formats -- This is possible.  I'd suggest that it would be more the bit depth over the sampling frequency.  But most people can't hear the difference in quality beyond 16/44 (so I guess the attitude would be something like 'we provide sufficiently high quality audio to please 99% of the population; there's no benefit for us in increasing costs/complexity to please that residual 1% and thus hit market share / technology adoption rates')

Part of the problem (for many people, not necessarily you) could be the actual sound of the hi-fi (as it were).  I gave up worrying about recorded sound quality when I started going to live performances (well, recitals, that sort of thing, not pop concerts) -- the actual sound of instruments playing is so far removed from that which you'd get when playing a recording.

That final paragraph sums up my attitude to music in a nutshell

A live concert provided the musicians are actually playing the instruments always provides an experience that a recording can never really replicate. I even enjoy the odd bum note and duff gig because it at least adds an element of danger to the performance.

It's amazing how some popular musicians simply can't cut it live. Makes me wonder if they actually played on their recordings at all. By contrast I have seen semi pro bands playing in bars and pubs who can provide an audience with a great evening out. A hi fi system no matter how good is never going to match the shared experience of going to a decent gig.

[Scunnered]

13 hours ago, spyguy said:

Even now, my 10 yo goes  can you hear that high pitch whine. Err no.

<bernardmanning>

Is it ... the wife?

</bernardmanning>

[DTMark]

3 hours ago, Craig_ said:

As a vinyl fiend I'd always been a bit sniffy about digital audio in general, but listening to Spotify on extreme quality (320kbps I think) through my B&W car stereo is probably the finest listening experience I've ever had. What does bug me with a lot of modern music is the absurd compression which is applied, makes everything sound like it was recorded in a toilet.

I realise this doesn't add much to the argument, my suspicion is that the OP has particularly highly-functioning ears, possibly honed by the ingestion of Class A's...

< Giggles

It's not that. I've always had this right from when CD was "invented" through to today.

FWIW 320k MP3 is about level with a ferric or maybe chrome tape cassette to me. It's really very compressed and lacking a lot of detail. YouTube, at its best, is 320k MP3. Take a good sounding recording from YouTube then try the same on Tidal at 16/44. The 16/44 will blow the 320K one away to such an extent that they barely sound like the same music. Mind you, in a car, 16/44 is more than acceptable to me. It's only obvious when I play them through the main system.

The seemingly obvious explanation is that my hearing is "better" than other peoples' hearing.

Except that I know that I can only hear to 16k. I've tested myself. Probably not too far from normal for my age, actually maybe a little sub-standard. Partner can hear to 18k.

To be fair, I'm far from alone in how I hear CD, though I'm in a small minority of people who seem to hear the compromises made and consider them unacceptable.

To clarify at this point: vinyl is riddled with all sorts of distortion and colouration. This is not about "vinyl is better", it all comes down to which set of distortions and colourations you can deal with. I'd prefer that the midrange bloat of vinyl (on my TT, anyway) was tempered, the bass was faster, and the top end more incisive. Like CD, in fact. Just not distorted.

To pick up on the points above and my own surprising result:

I recently spent some time with a modern DAC, a Meridian Explorer 2. It's a budget thing, only circa £200. Selected because it has MQA. More on that later.

16/44 remains unlistenable. It is choppy, truncated, sounds like it's stumbling from bit to bit in a jagged fashion and is about level with a good cassette deck overall. Better in some ways and worse in others.

24/92 is audibly and distinctively different. I can pick it apart from 16/44 easily though it still doesn't do much for me and I can't always tell which way around they are in a blind test because they're both "below par" and a bit, well, crappy.

It takes 24/192 for digital to begin to be acceptable to my ears. I can pick that one out every time. The bit rate seems to be concerned with detail and the sampling rate with timing. Which makes some sense because of the way in which band limiting works. If you remove a set of frequencies you deliberately introduce distortion, said to be inaudible, and if the listener can detect that "gap" then you have timing failure (and frequency failure too, both sides of the same coin).

For the sample track that I have - I have that on vinyl both on the LP and on the B side of a 45 RPM 12" single. The track was an analog recording.

Digital is tighter, less bloaty, crisper and more incisive. The LP is poorly pressed/mastered. The 12" has the obvious benefit that it's spinning faster and massively outperforms the 33RPM album.

The 45RPM vinyl beats the 16/44 but it doesn't beat the 24/192. Different strengths and weaknesses, but they're about level.

So for me 24/192 is the "starting point" for digital. Now, where's the 240/1920 copy..

But this rather blows away the theory that you won't hear more if you up the sampling and the bit rates.

CD was designed with the principle in mind that since humans can't hear beyond 22.5k (most can't hear beyond 20k) then you can discard that data and hear no difference making it possible to turn an infinite analog stream into a bit stream with a manageable file size. It is impossible to hear that something has been removed. You will never know.

Since then research has shown that humans do indeed respond to ultrasonics up to and beyond 30k. The brain registers activity even though the ears "hear" nothing.

Ok, it's the missing ultrasonics that are the issue. It is that simple. 24/192 is the threshold. It's the minimum you need for human hearing. Or should I say, my hearing.

Except..

This rather presumes that if I play vinyl I hear ultrasonics. Vinyl isn't band-limited and could indeed preserve them, perhaps in optimal circumstances. For example, it needs to be mastered from something better than a 16/44 copy (this is audible to me with vinyl, I can spot something mastered from low-res digital). Use the CD as the master for vinyl and all the ultrasonics have gone anyway. But if they are present: can the speakers actually reproduce them? Most tail away beyond 20k.

Clearly, there's something going on with human "hearing" here. Not everyone is the same. After all some are short or long sighted, some colour blind, and we never really know for sure that what the person next to us sees, is exactly the same as what we see.

For someone "into" hi-fi, the last 25 years or so have been so frustrating. I try 16/44 over and over, and every time, it sounds as gruesome as it ever did. FWIW another member on here did manage to acquire a CD player at some cost which overcame these issues to some extent. And to be fair I have quite a good turntable that in today's money would cost a lot more than the majority of CD players.

However: along comes "MQA". For those who haven't read about this, it's a technology designed to "fix" digital. It asserts that digital is "failed" (largely due to timing) and aims to redress this. It aims to remove a particular type of digital artefacting called "ringing" as much as possible (a consequence of band limiting and digital generally) and is an end to end process from the studio to the DAC and as such needs an MQA DAC to get the most from it.

That's the DAC I bought so I was able to try this, and it does indeed work. MQA is better than 16/44, an inconvenient truth that seems to have shaken the digital purist community. In my brief testing I thought it about level with 24/192 but then the DAC broke and I sent it back.

If I have what you might call "special hearing" - and this is one theory I have refused to accept thus far because there is no real evidence for it beyond this, maybe this does explain it. It is that simple.

However I'm intrigued as to whether others can hear the same so if you want to try it yourself (you'll need something capable of playing 24/192 natively) do let me know and I can make the files available.

[DTMark]

4 hours ago, stormymonday_2011 said:

That final paragraph sums up my attitude to music in a nutshell

A live concert provided the musicians are actually playing the instruments always provides an experience that a recording can never really replicate. I even enjoy the odd bum note and duff gig because it at least adds an element of danger to the performance.

It's amazing how some popular musicians simply can't cut it live. Makes me wonder if they actually played on their recordings at all. By contrast I have seen semi pro bands playing in bars and pubs who can provide an audience with a great evening out. A hi fi system no matter how good is never going to match the shared experience of going to a decent gig.

This is backed up by what I have read.

You might think that a cellist, for example, would have a top end hi-fi system and be extremely critical of audio reproduction.

That appears not to be the case.

Largely because the hi-fi, whatever it is, doesn't really reproduce the sound of the cellist accurately, it's just a "snapshot" and doesn't really get close to what a cellist would sound like were they in the room playing in front of you.

[MrPin]

For some reason I do enjoy my vinyl Some of it is a bit crackly.

My recent CD player seems to avoid the tinniness of the previous model.

[DTMark]

7 minutes ago, MrPin said:

For some reason I do enjoy my vinyl Some of it is a bit crackly.

My recent CD player seems to avoid the tinniness of the previous model.

Much of my vinyl crackles (the more recent it is, the more it seems to crackle).

It's a total no-go as a format for classical and other music with quiet bits.

Thankfully I don't listen to that type of music so often. However I wouldn't buy classical on vinyl. I do have some classical stuff and it sounds incredible, until it, er, crackles. That I can pick out the location of the performers - and maybe even height (a couple of classical LPs seem to convey the impossible with stereo!) it all for nothing when it cracks and pops.

Given that I have dipped into CD over the years.. and from my limited experience..

Some brands tend to avoid that top end thinness. Marantz and Sony seem to go for a sonorous rich warm sound. One Marantz CD player I heard was so warm sounding it made me chuckle thinking of people who consider CD to be "accurate" and vinyl "warm". My turntable sounds incredibly clinical by comparison.

Some CD players from the early 2000s seem to have that "it will scrape the paint off the walls" style of top end.

The DAC I tried most recently actually sounded a bit like the top end had been curtailed slightly, a little muted and flat, thus complementing 16/44 and trying to squash that "glassiness" to the top end albeit without success. However it was never "harsh" sounding in the way that earlier tech could be.

[spyguy]

3 hours ago, MrPin said:

For some reason I do enjoy my vinyl Some of it is a bit crackly.

My recent CD player seems to avoid the tinniness of the previous model.

A high end DSP, feeding into a high res DAC, connected to good speakers on a line protected from the PSU noise, can make the music sound like a wax disc, shellac, vinyl, cd, live gig.

Seriously, you people are all precious flowers!

hove an osciloscope on the output line, sample from a mike.

Unless you're playing something on a crap.player through crap noisy speakers ..

 

[MrPin]

Yes Mr Mark. I bought a Marantz and it does CD, and SACD fine. SA8500. Rather heavy. Older Marantz never had quite the sound, but I suppose CDs were newish then.

[Deft]

I think the general jist is that 16-bit is about enough dynamic range compared to the human ear, or sufficiently far enough away from the noise floor of most equipment to be good enough for recording. In terms of sample rate, Nyquist has never been proved "wrong" in the theory of signal sampling / reconstruction.

I think generally there are a lot of other factors that can come into play, filters, dithering, strange quirks of audio interfaces / chipsets / hardware. Perceived wisdom was there is maybe something to be gained by using 24-bit for live recording, but higher sample rates typically viewed as a waste. Things might have moved on. Hang-ups about 16-bit, like someone mentioned could be more to do with the first tranch of cd players etc.

Given that MP3 has won the war, it's not worth worrying too much about it.

Hugh Robjohns and co at Sound on Sound always had good articles going through some of this stuff, debunking some of the stuff around the perceived "stepped" nature of digital audio etc.

[Habeas Domus]

I'm a big fan of HDCD, which squeezes a 20 bit depth onto a CD, I agree with the Nyquist theory and I don't understand why it sounds better, it just does.

https://en.wikipedia.org/wiki/High_Definition_Compatible_Digital

Unfortunately there arent a lot of CDs or CD players that support HDCD, Joni Mitchell's back catalogue has it.

This does make me think that higher resolution files are the way to go, I think ideally there is a sweet spot somewhere between 16/48 and 24/96

[Scunnered]

On 15/10/2016 at 3:13 AM, DTMark said:

I can pick out 24/192 (hi-res) vs 16/44 (CD) every time with a 100% success rate be it on my computer, or through the main system with a modern DAC. It's dead easy. That horrible distortion at the top end and that awful choppy sense of truncation has gone. It's a total give-away.

Does this involve a particular type of music, or can it be anything?  I can imagine that electronic music, for example, might interact with the digitisation process in ways that music produced by acoustic instruments wouldn't.

[frozen_out]

Maybe you're an android?

Maybe you have confirmatiom bias and in a true blind test under laboratory conditions you wouldn't be able to hear things that are technically inaudible.

[John51]

As sample rates get higher and higher, some audiophiles keep tweaking the 16 bit dacs and some are very very good these days.

Have you listened to a non oversampling dac?

When the CD was first developed, the engineers didn't like the 'steps' involved in reproducing the waveform. So they over sampled to smooth out those steps. Kinda like household stairs being filled in with Lego bricks to give loads of smaller steps and a smoother looking slope.

As it turns out, listeners can find oversampling fatiguing due to the micro changes in rhythm, which we are sensitive to. Gives the 'glassy' sound. In practice, having the bigger 'steps' isn't that bad as our brain can easily fill in the gaps. 

I have a Devilsound NOS dac, cost £300 5 years ago and it's great. It's about 70% there on 'sounds like analogue'. I get no listening fatigue from this dac. 

Find a hifi store that carries a good NOS dac and take your reference CD.